Darolutamide Associated With Less Cognitive Decline Than Enzalutamide in Advanced Prostate Cancer

Darolutamide (Nubeqa) produced significantly less cognitive decline than enzalutamide (Xtandi) over 24 weeks in men with advanced prostate cancer, according to results from the phase 2 ARACOG trial (AFT-47) scheduled to be presented at the 2026 ASCO Annual Meeting.^1,2

The findings represent the first randomized, prospective head-to-head comparison of the cognitive effects of two androgen receptor pathway inhibitors (ARPIs) in a US population, filling a critical evidence gap in a disease setting where quality of life is a primary treatment consideration.

"This is the first randomized comparison of American patients receiving enzalutamide or darolutamide for advanced prostate cancer that compares cognitive effects as the primary endpoint. Enzalutamide and darolutamide appear to work similarly in terms of prostate cancer control. But knowing that there can be differences in cognitive effects between these drugs may affect a clinician's choice for treating prostate cancer if both options are available," said lead study author Alicia Morgans, MD, MPH, of Dana-Farber Cancer Institute in Boston.

ARACOG Trial Rationale and Study Design

Although both darolutamide and enzalutamide are approved for multiple advanced prostate cancer populations and have demonstrated significant improvements in overall survival across disease states—including nonmetastatic castration-resistant prostate cancer (nmCRPC), mCRPC, and metastatic hormone-sensitive prostate cancer (mHSPC)—prior data have suggested meaningful differences in their central nervous system effects. Preclinical studies have shown substantially lower blood-brain barrier penetration with darolutamide compared with enzalutamide, a difference rooted in their distinct molecular structures.

According, investigators launched ARACOG (AFT-47), a randomized phase 2 study that enrolled 111 patients with mCRPC, nmCRPC, or mHSPC at US academic medical centers and affiliated satellite sites between August 17, 2021, and March 11, 2025. Patients were randomized 1:1 to receive darolutamide (n = 55) or enzalutamide (n = 56) and were stratified by age (younger than 65, 65-80, and older than 80 years).

The primary end point was the percentage change in Maximally Changed Cognitive Domain (MCCD) between baseline and 24 weeks, assessed using five remotely deliverable Cambridge Neuropsychological Test Automated Battery (CANTAB) modules covering executive function, visual memory, attention, and working memory.

Secondary assessments included patient-reported outcome measures—FACT-Cog (perceived cognitive function), FACT-P, PHQ-9, and a sleep scale—as well as the Timed Up and Go (TUG) functional assessment and adverse events of interest including falls, severe neurologic toxicity, posterior reversible encephalopathy syndrome (PRES), and seizure.

Patients could cross over to the other treatment at 12 or 24 weeks if they met any one of four prespecified criteria: a ≥30% decline in any CANTAB module, a ≥10-point decline in FACT-Cog, a fall or investigator-assessed increased fall risk, or a grade 2 or higher neurologic toxicity event. Crossover could also occur outside of the 12- and 24-week assessment windows if a severe neurocognitive adverse event occurred. Patients with a fall, seizure, or PRES on darolutamide would come off study to avoid the risk of similar events if crossing to enzalutamide.

Patient Characteristics

Of the 111 enrolled patients, 95 were evaluable for the primary endpoint: 48 in the darolutamide arm and 47 in the enzalutamide arm. Baseline characteristics were broadly balanced between arms. Median age was 70.7 years (IQR, 48.6-86.2) in the darolutamide arm and 71.3 years (IQR, 50.0-89.5) in the enzalutamide arm. Median PSA was 1.0 ng/mL (IQR, 0.1-3.9) and 2.2 ng/mL (IQR, 0.2-8.3), respectively. The most common disease state at enrollment was mHSPC (54.5% darolutamide; 50.0% enzalutamide), followed by mCRPC (34.5% and 44.6%) and nmCRPC (10.9% and 5.4%).

The population was predominantly White (92.7% darolutamide; 73.2% enzalutamide), with 1.8% and 12.5% identifying as Black or African American/African Heritage, and 5.5% and 14.5% identifying as other race or not reported, respectively—a racial imbalance investigators identified as a study limitation, noting that most Black patients and those reporting other race were randomized to the enzalutamide arm.

Prior chemotherapy was infrequent in both arms (9.1% darolutamide; 10.7% enzalutamide), with a median time from prior chemotherapy to enrollment of 36.6 months in the darolutamide arm and 32.7 months in the enzalutamide arm. Education levels were comparable between arms. No patients in the darolutamide arm and 1.9% in the enzalutamide arm had left school before age 16. Among those with available education data, 28.6% of darolutamide patients and 17.3% of enzalutamide patients had left school at ages 17-18; 57.1% and 67.3% had completed undergraduate education; 8.2% and 9.6% held a master's degree; and 6.1% and 3.8% held a professional degree, respectively.

Primary Outcome: MCCD at 24 Weeks

Treatment with enzalutamide was associated with a significantly greater decline in objectively measured cognitive function than darolutamide over the 24-week observation period. The MCCD—the test showing the greatest degree of change for each patient—was the PALFAM test (assessing visual memory and executive function) for the darolutamide arm, with an 80% contribution from visual memory and 20% from executive function. In the enzalutamide arm, the MCCD was the SWM test (assessing working memory and executive function), with equal 50%/50% contributions. The median change in MCCD score was -15.8% in the darolutamide arm compared with -36.1% in the enzalutamide arm (P =.009), representing a statistically significant and clinically meaningful difference in cognitive decline between the two agents. Patients who crossed over prior to 24 weeks were analyzed using their cognitive score at the time of crossover in their randomized treatment arm.

The pattern of cognitive change also differed qualitatively between the two arms. Participants in the darolutamide group appeared to improve at testing over time—consistent with a learning effect, in which repeated exposure to cognitive testing leads to incremental performance gains—whereas participants in the enzalutamide group did not demonstrate this expected learning effect, suggesting that enzalutamide-related cognitive impairment was actively counteracting the practice benefit.

Secondary End Point: Crossover

Although a similar number of patients in each arm met the protocol-defined criteria to switch treatments—36 unique patients eligible in the enzalutamide arm and 34 in the darolutamide arm—only patients receiving enzalutamide actually chose to cross over. A total of 30 enzalutamide-treated patients crossed over to darolutamide during the trial: 2 between baseline and 12 weeks, 16 at the 12-week assessment, and 12 at 24 weeks. Among the 36 patients eligible to cross over from enzalutamide, 17 were uniquely eligible at 24 weeks, of whom 14 had previously met crossover criteria at 12 weeks. In the darolutamide arm, despite 34 patients meeting eligibility criteria for crossover—including 24 at the 12-week timepoint and 24 at 24 weeks, with 14 previously meeting criteria at 12 weeks—zero patients chose to switch to enzalutamide at any time point throughout the trial.

The most common reasons for crossover from enzalutamide were worsening cognitive impairment documented by testing or reported by the patient. Investigators noted that patients in the darolutamide group may have been less bothered by their symptoms, and also acknowledged that the cost difference—darolutamide provided at no copay versus enzalutamide requiring a copay—may have influenced the decision not to switch.

Limitations and Next Steps

The investigators acknowledged several study limitations. Because enzalutamide was provided through standard of care, the requirement for acceptable copayment amounts may have introduced selection bias and influenced crossover rates. The trial was conducted exclusively at academic medical centers in the US, with regional and satellite sites covered by academic center IRBs, potentially limiting generalizability to community practice settings. The predominantly White patient population—and the racial imbalance between arms, with most Black patients and those reporting other race randomized to enzalutamide—limits the applicability of findings across racially diverse populations.

Researchers are continuing to follow participants to evaluate the drugs' cognitive effects out to 48 weeks and are also investigating whether genetic factors, including polygenic hazard scores and androgen receptor testing, may contribute to individual susceptibility to cognitive change with ARPI therapy.

"Darolutamide and enzalutamide are widely used and highly effective treatments for advanced prostate cancer, but they may differ in their impact on cognitive function. In a new randomized phase 2 study, men treated with darolutamide experienced less cognitive decline than those receiving enzalutamide. This difference may be explained by darolutamide's limited ability to enter the brain, potentially offering a favorable option for preserving cognitive health in men with advanced prostate cancer," Samuel U. Takvorian, MD, MSHP, Deputy Director of the Penn Center for Cancer Care Innovation at Abramson Cancer Center and an ASCO Expert in genitourinary cancers said in a statement.²

The study was funded by a Prostate Cancer Foundation Challenge Award, an Alliance Scholar Award, and Bayer.

REFERENCES

1. Morgans AK, et al. ARACOG (AFT-47): A randomized phase II study of androgen receptor directed therapy on cognitive function in patients treated with darolutamide or enzalutamide. J Clin Oncol 44, 2026 (suppl 16; abstr 5005).

2. American Society of Clinical Oncology. Less cognitive decline with darolutamide treatment than enzalutamide in people with advanced prostate cancer. ASCO press release. May 21, 2026. Embargoed for release May 21, 2026, at 5 PM ET.